DeepSeek/awesome-deepseek-integration
1
0
Fork 0
mirror of https://github.com/deepseek-ai/awesome-deepseek-integration synced 2025-01-22 10:35:53 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

docs(huixiangdou): add readme

Browse Source
This commit is contained in:
tpoisonooo 2024-02-02 20:01:43 +08:00
parent 11bcb15feb
commit a6e18c72bb
2 changed files with 93 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

53
docs/huixiangdou/README.md Normal file
View File

@ -0,0 +1,53 @@
<img src="https://github.com/InternLM/HuixiangDou/releases/download/v0.1.0rc1/huixiangdou.jpg" width="64" height="auto" />
# [HuixiangDou](https://github.com/InternLM/HuixiangDou)
"HuixiangDou" is a domain knowledge assistant based on LLM. Features:
1. Handles complex scenarios such as group chat and handle answer user questions without flooding messages.
2. Proposes a solution for technical issues through an algorithm pipeline.
3. Low cost, see [arxiv2401.08772](https://arxiv.org/abs/2401.08772)
# Demonstration
HuixiangDou is a **assistant**
* Says it doesn't know in response to halluciations
* Personal Wechat supported
* Feishu supported
<img src="https://github.com/InternLM/HuixiangDou/releases/download/v0.1.0rc1/demo0.jpg" width="400">
HuixiangDou is a **group** assistant
* Ignores idle talk
* Only interacts with those who need help
<img src="https://github.com/InternLM/HuixiangDou/releases/download/v0.1.0rc1/inside-mmpose.jpg" width="400">
HuixiangDou is a **domain knowledge** group assistant
* Relies on an algorithm pipeline + LLM instead of training
* Has faced thousands of users, and has been "teased" over 20,000 times in half a year
<img src="https://github.com/InternLM/HuixiangDou/releases/download/v0.1.0rc1/inside-middleware.png" width="400">
# Configuring deepseek API
**STEP1.** Apply for a token at deepseek open platform
**STEP2.** Follow the HuixiangDou Advanced Configuration and fill in the config.ini with your token
```ini
# config.ini
[llm]
enable_local = 0
enable_remote = 1
..
[llm.server]
..
remote_type = "deepseek"
remote_api_key = "YOUR-API-KEY"
remote_llm_max_text_length = 16000
remote_llm_model = "deepseek-chat"
```
**STEP3.** Run
```shell
python3 -m huixiangdou.main --standalone
```

42
docs/huixiangdou/README_cn.md
View File

@ -12,10 +12,48 @@
## 效果展示
1. 它是个**机器人**
* 面对幻觉会说不知道
*
* 支持微信
* 支持飞书
![image](https://github.com/deepseek-ai/awesome-deepseek-integration/assets/59196087/367a41b6-3277-4897-a53f-aaa17bb9dc53)
<img src="https://github.com/InternLM/HuixiangDou/releases/download/v0.1.0rc1/demo0.jpg" width="400">
2. 它是个**群组**机器人
* 不会搭理闲聊
* 只找需要帮助的人
<img src="https://github.com/InternLM/HuixiangDou/releases/download/v0.1.0rc1/inside-mmpose.jpg" width="400">
2. 它是个**领域知识**群组机器人
* 不依赖训练，依靠算法 pipeline + LLM
* 已经面对数千用户，半年被“调戏”了 2w 次以上
<img src="https://github.com/InternLM/HuixiangDou/releases/download/v0.1.0rc1/inside-middleware.png" width="400">
## 配置 deepseek API
**STEP1.** 打开 [deepseek 开放平台](https://platform.deepseek.com) 申请 token
**STEP2.** 参照 [HuixiangDou 高级配置](https://github.com/InternLM/HuixiangDou?tab=readme-ov-file#step4-advanced-version-optional)，把 token 填入 `config.ini`
```ini
# config.ini
[llm]
enable_local = 0
enable_remote = 1
..
[llm.server]
..
remote_type = "deepseek"
remote_api_key = "YOUR-API-KEY"
remote_llm_max_text_length = 16000
remote_llm_model = "deepseek-chat"
```
**STEP3.** 运行
```shell
python3 -m huixiangdou.main --standalone
```